ON THE RELATIONSHIP BETWEEN LIGHTNING AND CONVECTIVE RAINFALL
OVER THE CENTRAL PACIFIC OCEAN
Antti Pessi, Steven Businger
University of Hawaii
Honolulu, Hawaii, United States
Kenneth L. Cummins, Tommy Turner
Vaisala Inc.
Tucson, Arizona, United States
1. INTRODUCTION
The wave guide between the Ionosphere and the Earth's surface allows VLF emissions generated by lightning between 5 and 25 kHz (sferics) to propagate over very long distances. The goal of the new Pacific Lightning Detection Network (Pacnet) is to collect long-range lightning data over the central and north Pacific Ocean and investigate the relationship between lightning and rainfall rates and their implications for storm evolution and modelling. Four hybrid (IMPACT) broadband receivers have been installed on Unalaska, Kauai (Lihue), Island of Hawaii (Kona) and
Kwajalein (Fig. 1). A fifth site in Kiritimati is planned, and others may be added in the future. This paper presents our first (preliminary) observations with this network.
Pacnet lightning detectors have a hybrid design, which includes some of the advantages of accuracy at close range of traditional broad-band detectors and the advantages of long range associated with VLF detectors. The sensors use combined technology which utilizes both time-ofarrival and direction finding methods in the data processing. Because of its long-range detection capability, Pacnet provides inexpensive, accurate monitoring of convective storms that represent a threat to ocean shipping and airborne carriers and are beyond the range of weather radars.
Although GOES and GMS provide continuous IR imagery, convective activity is often obscured by cirrus anvils. Some of the cirrus-covered clouds pose little hazard to aircraft operations, whereas cold clouds that produce lightning have significant updrafts, increasing the threat of turbulence and icing. Low orbiting satellites that carry microwave radiometers such as NASA's EOS-Aqua with its
AMSR-E (Advanced Microwave Scanning Radiometer - Earth Observing System) and Tropical
Rainfall Measuring Mission (TRMM) with its TMI (TRMM Microwave Imager), provide intermittent glimpses of convective precipitation. Unfortunately, they do not permit the evolution of convective weather systems to be monitored continuously from space.
The distribution of moisture, and in particular, latent heating in deep convection, is critical for accurate forecasts of cyclogenesis. A promising application of Pacnet is to derive estimates of the rainfall rate from the lightning data. The relationship between lightning and rainfall may vary significantly, depending on air-mass characteristics and cloud microphysics. Several studies have concentrated on the mid-latitude continental storms. The nature of the relationship is less certain over the oceans and in particular, over the tropical oceans (Petersen and Rutledge, 1998; Zipser,
1994). Alexander et al. (1999) found relatively good correlation between convective rainfall and lightning rates on 1993 Superstorm and improved numerical forecast by assimilating latent heating rates derived from lightning data. They used SSM/I data to derive rainfall rates and lightning observations from both NLDN and VLF networks. Chang et al. (2001) used VLF sferics observations and TRMM satellite data to derive the relationship between convective rainfall and lightning on 1998 Groundhog Day Storm. They also assimilated the latent heating profiles and improved forecasts.
Our hypothesis is that in cases of cyclogenesis in marine air masses, including subtopical cyclogenesis, the relationship between rainfall and lightning rates will be relatively robust because of a more uniform aerosol and cloud microphysical environment. Prelimary results of a comparison of the lightning rate measured by Pacnet and convective rainfall obtained from Aqua's and TRMM's microwave sensors for a variety of storm systems over the central north Pacific are presented. The

results indicate that the ratio of lightning to rainfall rate shows a relatively stable relationship over the
Pacific Ocean, suggesting that lightning data over the Pacific can be assimilated into numerical models as a proxy for latent heat release in deep convective clouds.
Fig. 1 Map of the central Pacific Ocean showing locations of Pacnet sites: the red, solid triangles are installed, whereas open triangles are potential future sites. Also shown is a subset of the
International Long-Range Network (blue triangles: Japanese, and green triangles: North American sites).
2. METHODOLOGY
Pacnet sensors are located in the central Pacific with contributing ILRN (International Long-
Range Network) sensors in North America and Japan (Fig. 1). In combination they can detect lightning over the whole northern Pacific with relatively good location accuracy (median values of 4-
8 km near Hawaii and falling off to 26-32 km in the mid-pacific). Detection efficiency of the network over the north Pacific was increased significantly (30-40%) when Pacnet sensors were added to the existing ILRN. Detection efficiency depends on the time of the day, being best during the night and less good during the day. In this study, a detection efficiency correction coefficient for night/day cases is not used since it is not well known yet. All the sensors use combined technology which utilizes both time-of-arrival and direction finding methods in the data processing. Lightning data for the case studies were obtained from Pacnet or ILRN, depending on data availability.

3. RESULTS
Two cases were analyzed using the available long-range lightning data.
3.1 Subtropical low over Hawaii
A subtropical cyclone (kona low) impacted the Islands of Hawaii during the period 26-29
February 2004, causing heavy rain, thunderstorms, extensive flooding and two storm-related deaths. The MM5 6 hour precipitation forecast shows a squall line over Oahu and Molokai (Fig. 2).
Lightning and satellite observations (Figs. 3 and 4) place the rainband ~200 km to the east. We suggest that the addition of lightning diagnosed rainfall in the initial condition for MM5 may have resulted in a more accurate forecast, a subject of current research.

Fig. 2. MM5 6 hour surface pressure and rainfall forecast initialized at 00 UTC on 28 February
2004.
The number of lightning observations over Hawaii increased significantly when Pacnet sensors were enabled (Fig. 3). Most of the lightning over Hawaii was located by the two local sensors in
Lihue and Kona.

Fig. 3. Lightning strokes near Hawaii on Feb 28, 2004 00-12 UTC (a) Without Pacnet sensors enabled and (b) with Pacnet sensors
The main rainband was located on the east coast of the Big Island at the time of Aqua's overpass at
~12 UTC (Fig. 4a). As the squall line moved southeastward, less lightning strikes were observed on the southeast side of the Big Island than expected (Fig. 4b). This observation is the consequence of this region falling along the exterior baseline between the Kona and Lihue sensors, where the twosensor network geometry is poor.
Fig. 4. (a) Convective rainfall rate (mm/h) from Aqua's AMSR-E sensor at ~12:00 UTC Feb. 28,
2004. (b) Lightning strokes from Pacnet ~11:50-12:10 UTC.
3.2 Extratropical storm on the northeast Pacific
Beginning on 16 December 2002, an extratropical cyclone produced a large amount of lightning as it moved across the east Pacific Ocean. On December 18, the center of the storm was located near 43N 135W, northeast of Hawaii and west of Oregon. Most of the lightning activity was associated with the cold front (Fig. 5).

Fig. 5. (a) Satellite image at 18 UTC on 18 December 2002 and (b) 03 UTC December 19 (right).
ILRN Lightning detected between 1453 and 1753 UTC and 2353 and 0253 UTC December 18-19
(right) . Red dots indicate lightning during the last hour and yellow dots during the previous two hours.
A geographic correlation between lightning occurrence and convective rainfall is evident in
Fig. 6.
Fig. 6. (a) Convective rainfall derived from TMI sensor data for one satellite overpass taken at
~21:45Z. (b) Lightning strokes from ILRN within 30 min of satellite overpass time.

The shape and location of the cold front was relatively well observed by ILRN (Fig. 7)
Details of the convective pattern are mirrored in the lightning rate. Additionally, weak convective rainfall was observed along the southern part of the front but no lightning, suggesting that either the convective activity was too weak there to produce lightning, or that the detection efficieicy of the
ILRN (no Pacnet sensors) was falling off at larger distances from the continental U.S.
As mentioned before, the detection efficiency depends on the time of the day. In this case the local time of the satellite overpass was close to mid-day when detection efficiency was at its poorest.

Fig. 7. TMI convective rainfall (a). ILRN lightning strokes (b). These figures are close-ups of Fig. 6.
The relationship between the lightning and rainfall rates was investigated using a probability matching technique that described in Calheiros and Zawadski (1987). Convective rainfall rates up to
4 mm/h were observed, corresponding to a lightning rate of 36 strokes/30 minutes. The resulting relationship over the northeastern Pacific Ocean is reasonably robust (Fig. 8), providing encouragement for the goal of using lightning rates as a proxy for rainfall in numerical models over the north central Pacific.
Fig. 8. Convective rainfall rate vs. sferics rate observed in the northeast Pacific storm 18 December
2002.
4. SUMMARY AND CONCLUSIONS
The goal of the new Pacific Lightning Detection Network (Pacnet) is to provide long-range lightning detection over the central and northern Pacific Ocean for research and operational applications.
Four lightning detection sensors have been installed in the central Pacific. Data from the new hybrid sensors significantly improved detection efficiency and location accuracy in the central and north Pacific, when compared to data from the existing long-range network. Lightning data from Pacnet and ILRN and rainfall data from microwave sensors aboard low-orbiting satellites were used to study the relationship between lightning and convective rainfall rates over the Pacific.
Because of its long-range detection capability, Pacnet provides inexpensive, continuous monitoring of convective storms that represent a threat to ocean shipping and airborne carriers and are beyond the range of weather radars. Although GOES and GMS provide continuous IR imagery, convective activity is often obscured by cirrus anvils. Some low orbiting satellites can detect lightning.
Unfortunately, they do not permit the evolution of convective weather systems to be monitored

continuously from space.
The distribution of deep moisture, and in particular, latent heating in deep convection, is critical for accurate forecasts of cyclogenesis. A promising application of Pacnet is to derive estimates of the rainfall rate from the lightning data. In this study the ratio of rainfall to lightning over the Pacific is investigated by comparing the number of lightning strokes measured by Pacnet and convective rainfall obtained from TRMM's and Aqua’s microwave sensors for a variety of storm systems. The relationship between lightning and rainfall may vary significantly, depending on airmass characteristics and cloud microphysics. Preliminary results indicate that the ratio of lightning to rainfall rate shows a stable pattern over the Pacific Ocean, consistent with the uniform aerosol and cloud microphysical environment over this region. This result holds promise that lightning data over the Pacific can be assimilated into numerical models as a proxy for latent heat release in deep convective clouds. However, more case studies need to be completed to confirm this early promise.
REFERENCES :
Adler, R., T. Wilheit, Jr., C. Kummerow, and R. Ferraro. 2004. AMSR-E/Aqua L2B Global Swath
Rain Rate/Type GSFC Profiling Algorithm V001 , March 2004. Boulder, CO, USA: National Snow and Ice Data Center. Digital media.
Alexander, G. David, Weinman, James A., Karyampudi, V. Mohan, Olson, William S., Lee, A. C. L..
1999: The Effect of Assimilating Rain Rates Derived from Satellites and Lightning on Forecasts of the 1993 Superstorm. Monthly Weather Review : Vol. 127 , No. 7, pp. 1433-1457.
Calheiros, R.V., Zawadzki, I. 1987: Reflectivity-Rain Rate Relationships for Radar Hydrology in
Brazil .
Journal of Applied Meteorology : Vol. 26 , No. 1, pp. 118 –132.
Chang, Dong-Eon, Weinman, James A., Morales, Carlos A., Olson, William S. 2001: The Effect of
Spaceborne Microwave and Ground-Based Continuous Lightning Measurements on Forecasts of the 1998 Groundhog Day Storm. Monthly Weather Review: Vol. 129 , No. 8, pp. 1809-1833.
Kummerow, C., Y. Hong, W.S. Olson, S. Yang, R.F. Adler, J. McCollum, R.Ferraro, G. Petty, D.-B.
Shin, and T. T Wilheit, 2001: The evolution of the Goddard Profiling Algorithm (GPROF) for rainfall estimation from passive microwave sensors . J. Appl. Meteor ., 40 , 1801-1820
Petersen, W.A., and S.A. Rutledge, 1998: On the relationship between cloud-to-ground lightning and convective rainfall. J. Geophys. Res.
, 103 , 14025-14040.
Zipser, E. J.,1994: Deep cumulonimbus cloud systems in the tropics with and without lightning.
Mon. Wea. Rev ., 122 , 1837-1851.